Cadence Design Systems introduced Maestro on Monday, pairing AI models for chip and system design with agents and tools that customers can run together on their own infrastructure to address privacy and security needs.
The on-premises option applies to the complete environment: Cadence AI Super Agents, Maestro models and Cadence’s electronic design automation tools. Cadence said the package can also run on its cloud or secure public clouds, giving customers a choice of where the tools and data are handled.
That approach differs from one announced Sept. 30 by Synopsys and OpenAI. The companies plan to develop GPT-Synopsys under a multiyear partnership that includes revenue sharing, with the model set to run on OpenAI-hosted infrastructure. Both announcements concern AI for chip design, but specify different deployment options.
Maestro is designed to work with Cadence’s electronic and system design automation tools. Its agents can combine specialized models with leading frontier and open-weight models, then use feedback from design tools and deterministic solvers to evaluate engineering results, the company said.
NVIDIA Nemotron is among the models in the portfolio. Cadence said customers can further train models on their own designs, data and workflows. The company is also working to integrate NVIDIA’s OpenShell Reference Architecture for agent security.
Cadence reported that early design workflows generated results it rated as much as 30% better than those from a general-purpose foundation model. It said token efficiency in those tests was up to five times greater. The company did not specify the benchmark tasks, testing method or comparison model.
Cadence has not specified how many models will be in the first release or what they will cost. It also has not announced a date for availability beyond the early-access group.
Cadence said the first Maestro models are scheduled to reach select early-access customers in the fourth quarter of 2026.
